Study On Digital Image Auto-focusing And Restoration Technology Of The Defocused Image

Auto-focusing technology is a key technology in the imaging system and robot vision and various optical precision instruments. With the rapid development of science and technology, the auto-focusing problem has been much concerned by people, and the application in the field of image processing far and wide. Domestic and foreign scholars have made many researches and achievements in the digital image auto-focusing field. However, these researches focus on imaging system that are directed at non-real-time or on the CPU, digital signal processing (DSP). The computational complexity and the cost of implementation are higher. This paper seeks to provide a low-cost, high-speed real-time application of the ASIC algorithm.According to the character of the item, in a more comprehensive analysis of the clarity-focusing evaluations, we introduce a new arithmetic based on SML operator and associate with the CIS on chip system we build a hardware module, this module can be integrated on CIS chip. It is tested and worked very well on performance.According to the method of defocused restoration: the paper firstly introduces the defocused model in the traditional method: Pillbox model and Gauss model; infers the model can't reflect the factual defocused point spread function exactly, that results in unsatisfactory restoration of defocused image. In this paper the defocused point spread function can be assessed through knife-edge function curve, according to the circular symmetry of the defocused image. This method improves the estimating precision of the point spread function. Wiener filter is fabricated for recovering the defocused image based on the defocused point spread function. The impact of the noise to the restoration is reduced greatly by Wiener filter.The experiment indicates that with this module the speed and accuracy of the auto-focusing system have been improved greatly. And the improved method can assess the defocused point spread function more exactly than the traditional method, can gain clear image with higher resolution and contrast.